Challenges of Adversarial Image Augmentations
This work addresses the problem of optimizing image augmentation policies for machine learning practitioners, but it is incremental as it builds on and critiques existing methods like RandAugment and Adversarial AutoAugment.
The paper investigates whether random image augmentations remain competitive against optimal adversarial methods and simple curricula, finding that they do, and suggests that the success of Adversarial AutoAugment may stem from stochasticity in its policy controller rather than adversarial training itself.
Image augmentations applied during training are crucial for the generalization performance of image classifiers. Therefore, a large body of research has focused on finding the optimal augmentation policy for a given task. Yet, RandAugment [2], a simple random augmentation policy, has recently been shown to outperform existing sophisticated policies. Only Adversarial AutoAugment (AdvAA) [11], an approach based on the idea of adversarial training, has shown to be better than RandAugment. In this paper, we show that random augmentations are still competitive compared to an optimal adversarial approach, as well as to simple curricula, and conjecture that the success of AdvAA is due to the stochasticity of the policy controller network, which introduces a mild form of curriculum.